microsoft / onnxruntime

ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
https://onnxruntime.ai
MIT License
13.53k stars 2.76k forks source link

[Performance] Conv3D Support in QNN delegate #20532

Open kory opened 2 months ago

kory commented 2 months ago

Describe the issue

Conv3D support should be added to the QNN Delegate. Without it, 3D convs fall off to CPU. This affects models like ResNet3D--the entire model is not able to run on QNN because of this missing support.

To reproduce

See attached model.

Affects ResNet3D.

Urgency

No response

Platform

Linux

OS Version

Ubuntu 22.04

ONNX Runtime Installation

Built from Source

ONNX Runtime Version or Commit ID

Latest

ONNX Runtime API

Python

Architecture

ARM64

Execution Provider

QNN EP

Execution Provider Library Version

QNN 2.20

Model File

3d_conv_relu.onnx.zip=

Is this a quantized model?

No

HectorSVC commented 2 months ago

https://github.com/microsoft/onnxruntime/pull/20507